Metro 100G: Get Excited, but Not Too Excited
As part of Optical Transport Month here at Light Reading, I caught up with Heavy Reading senior analyst Sterling Perrin, who talked to me about a variety of pertinent optical issues. Here's Part 1 of our Q&A, in which Perrin analyzes the burgeoning metro 100G market, offering a tempered take on how the market the whole optical sector is talking about could actually play out.
Light Reading: Will 2015 be the year of 100G in the metro?
Sterling Perrin: In 2015, there will definitely be some very strong growth in 100G in the metro. It's already started. But I don't think it's going to be done this year. It's a couple-year rollout if you look at how it will work through North America and other regions. If you look at metro vs. long haul, the fastest growth rates for 100G in long haul are behind us, and for 2015 and 2016 and 2017, you'll see faster growth in the metro because that trend just hasn't hit yet. Metro's just kicking in now and the piece that's driving it -- which a lot of vendors have been saying, and there's definitely truth to it -- is the data center interconnect market moving from 10G to 100G. That's probably going to be the top driver to move to 100G in the metro.
Light Reading: Will we see 100G blanket all metros anytime soon?
Perrin: I don't expect there to be the same level of saturation of 100G in metro as we've seen in long haul. Basically the whole long haul market has moved over to 100G. In terms of capacity or ports shipped, it's massively in favor of 100G, and 10G is just dwindling in the long haul. I think 10G will have a long ride in metro because there are so many end points, and they're not all going to move to 100G. Some stuff is going to move from 1G and 2.5G to 10G. And some of the stuff that is 10G will move to 100G, particularly where you have a big capacity need for data center interconnect, but I don’t think it will take over the whole market the way it did in long haul.
Light Reading: Is there a little too much excitement over metro 100G data center interconnect right now?
Perrin: It does seem in general the market is getting a little more excited than it probably should be. Data centers are not everywhere. Look at Google's network as just one example. How many of these data centers do they have that are really the mega-scale data centers? Twelve or 13 globally? You can connect 12 or 13 sites globally with 100G and have a lot of 100G between them, but if you look at the typical telco and how many central offices they have, it's thousands. Those central offices need connection to other central offices and bigger central offices and also data centers, but those don't all move over to 100G by any means. When they do go to 100G, they will use a lot of it, but there is a lot of what they have that doesn't need 100G.
Light Reading: Do you think the vendor community understands the limits of the data center interconnect opportunity?
Perrin: On these conference calls, the vendors word things very carefully, and leave things open to interpretation. Internally what these vendors understand could be a bit of a different story. For example, if you look at Infinera, they have said a lot of great things about data center interconnect, and put out the CloudXpress to target that, but they also have talked about a second phase for their strategy, which is evolution of the legacy packet optical market. They've outlined that as a very separate piece. Data center interconnect is the faster growing market, and very open to new entrants, and it's very much a 100G opportunity, so it made sense for Infinera and BTI and Cisco, and all these other guys to go after it. But, if you're just looking at markets in total, where the greatest size is the packet optical market or legacy packet optical market, that continues to evolve and is not 100G for the most part.
Light Reading: With the long haul market largely converted to 100G, and the metro market just kicking in, where does the 400G evolution stand?
Perrin: We are kicking 400G out a little further at this point. OFC last year was really dominated with 400G and "beyond 100G" trials and demos, but the vendors have come to more of an understanding that it's not a real near-term revenue opportunity. You're still going to want to be positioned as a leader in R&D, and these types of experiments will still be going on, but I don't sense that many of these guys are betting significantly on it for growth. The general understanding is that 100G is the opportunity, and what needs to be done over the next five years is to really continue to bang costs out of 100G, in terms of the price you pay for a transponder, but also in terms of the size of 100G devices. The bulk of near-term R&D seems focused right now on bringing down costs and footprint on the line side.
If 100G is continuing to come down in costs, and 400G is just this bulky science experiment kind of box, there's no justification for deploying. Going from 10G to 100G, operators got a 10x capacity improvement, so in the long haul there isn't a real urgent need for 400G. When you do need more capacity, you can deploy more 100G. I think vendors have demonstrated 400G has a good understanding of where the priorities really are right now.
(Part 2 of our Q&A, focusing on IP and optical integration, will appear next week.)
— Dan O'Shea, Managing Editor, Light Reading